Multi-task intelligent seismic wave impedance inversion method and device

Abstract

The present application discloses a multi-task intelligent seismic wave impedance inversion method and device, the method comprising: performing well seismic calibration and determining target layer samples; obtaining density curves and velocity curves of logging data, and intercepting training sample labels from the density curves and velocity curves of logging data; obtaining seismic data of the same depth layer as the target layer from seismic data, and constructing a training data set for a velocity interpretation model; constructing a velocity interpretation model and a density interpretation model respectively, and determining the final density interpretation model P1 and the final velocity interpretation model S1; migrating the model parameters of model P1 and model S1 respectively, thereby constructing a wave impedance inversion model for multi-task learning, comprehensively training the wave impedance inversion model, and determining its loss function; and completing the inversion process. The present application improves the accuracy of seismic wave impedance inversion, has higher robustness, and improves vertical resolution.

Description

Technical Field

[0001] The present application relates to seismic wave impedance inversion technology, and in particular to a multi-task intelligent seismic wave impedance inversion method and device. Background Art

[0002] Seismic wave impedance inversion refers to the inversion of the wave impedance of underground strata by comprehensive use of seismic data and logging data with the help of computer technology. An effective reservoir wave impedance model provides a three-dimensional geological data basis for subsequent reservoir numerical simulation and plays a key role in oil and gas reserve calculations. Underground strata usually have multi-scale heterogeneity and continuity structures, but due to technical limitations, these parameters cannot be directly measured.

[0003] At present, geostatistical methods are mainly used for seismic impedance inversion. Seismic impedance inversion is a comprehensive reservoir characterization technology that combines seismic data, well logging data and core data. It aims to maximize the use of all collected data under the guidance of geological theory and express the geological information of the target layer in detail. However, in the process of seismic impedance inversion, the multi-solution of impedance data is still an urgent problem to be solved in reservoir modeling. Multi-solution makes it impossible to uniquely determine the formation attributes, thereby increasing the uncertainty of geological interpretation. Summary of the invention

[0004] The present application provides a multi-task intelligent seismic wave impedance inversion method and device to at least solve the aforementioned technical problems.

[0005] According to a first aspect of an embodiment of the present application, a multi-task intelligent seismic wave impedance inversion method is provided, comprising:

[0006] Conduct well seismic calibration and determine target layer samples;

[0007] Obtaining a density curve and a velocity curve of the well logging data, and extracting training sample labels from the density curve and the velocity curve of the well logging data;

[0008] Acquire seismic data of a layer segment with the same depth as the target layer segment from the seismic data, perform spectrum analysis on the acquired seismic data, use the seismic data after spectrum analysis and density data to construct a training data set for a density interpretation model, and use the seismic data after spectrum analysis and velocity data to construct a training data set for a velocity interpretation model;

[0009] A velocity interpretation model and a density interpretation model are constructed based on the training data set, and the seismic data are mapped to velocity and density according to the velocity interpretation model and the density interpretation model respectively; wherein the velocity interpretation model and the density interpretation model are both constructed using a gated recurrent unit (GRU) neural unit, and the number of GRU layers is 3; the dimension of the seismic data input to the velocity interpretation model and the density interpretation model is (B, 50, m), and the dimension of the output velocity and density is (B, 1, m), and B represents the number of samples of the model at each training;

[0010] The density explanation model is trained using the training data in the training data set of the density explanation model, as the final density explanation model P1 after the training is completed; the speed explanation model is trained using the training data in the training data set of the speed explanation model, as the final speed explanation model S1 after the training is completed;

[0011] The model parameters of model P1 and model S1 are migrated respectively, so as to construct a wave impedance inversion model of multi-task learning, wherein the wave impedance inversion model is a parallel combination of the density interpretation model and the velocity interpretation model;

[0012] The wave impedance inversion model is comprehensively trained and its loss function is determined; the wave impedance inversion model is used to complete the inversion processing of seismic data on density, velocity and wave impedance respectively.

[0013] In one embodiment, the method further comprises:

[0014] The velocity data and density data are obtained respectively according to the density curve and velocity curve of the logging data, and the product of velocity and density is calculated as the wave impedance.

[0015] In one possible implementation manner, the step of acquiring seismic data of a layer segment having the same depth as the target layer segment from the seismic data and performing spectrum analysis on the acquired seismic data includes:

[0016] Use the well coordinates to obtain the seismic trace data in the unit area grid centered at the corresponding position as the model input data, and use the following spectrum analysis method to obtain the frequency domain characteristics of the seismic data;

[0017]

[0018] in:

[0019] is a one-dimensional seismic trace signal, is the spectrum of the one-dimensional seismic signal in the frequency domain, is a plural unit, is the fundamental frequency of the one-dimensional seismic trace signal;

[0020] According to the spectrum characteristics, select the first main frequency and use the following bandpass filter , get the seismic waveform corresponding to the first main frequency:

[0021]

[0022] in, is the first main frequency, is the low-pass frequency value centered on the first main frequency, It is the high-pass frequency value centered on the first main frequency.

[0023] In one possible implementation, the constructing of a velocity explanation model and a density explanation model based on the training data set respectively includes:

[0024] Determine the number of target layer sample points as m, extract density and velocity training sample labels from the well logging data, the label dimension is (1, m), and obtain p×p grid seismic data of the same depth layer as the target layer from the seismic data, the dimension of which is (p, p, m);

[0025] The dimension of wave impedance is (1, m), and the seismic waveform corresponding to the first main frequency of seismic data is obtained by frequency filtering, and its dimension is (p, p, m);

[0026] Merge the seismic data and the seismic waveform corresponding to the main frequency f, merge the seismic data at different grid positions, and use the merged data as the data for the input model, with a dimension of (50, m).

[0027] In one possible implementation, the operation process of each layer of GRU in the three layers of GRU of the speed interpretation module and the density interpretation module includes the following processing steps: controlling how the state of the previous moment affects the current state; controlling how the current input affects the state of the previous moment; calculating the candidate hidden state based on the current input and the reset state of the previous moment; calculating the final hidden state at the current moment based on the impact of the state of the previous moment on the current state.

[0028] In one embodiment, the loss function of the wave impedance inversion model includes three parts: density loss , speed loss and wave impedance loss , calculated as follows:

[0029]

[0030]

[0031]

[0032]

[0033] in:

[0034] is the inversion loss of the density module;

[0035] is the inversion loss of the velocity module;

[0036] is the inversion loss of the wave impedance module;

[0037] N : The number of training samples;

[0038] : represents the true density value;

[0039] : represents the density value inverted by the density interpretation module;

[0040] : Indicates the true speed value;

[0041] : represents the velocity value inverted by the velocity interpretation module;

[0042] : represents the real wave impedance value;

[0043] : Represents the inversion value of the seismic wave impedance inversion model.

[0044] According to a second aspect of an embodiment of the present application, a multi-task intelligent seismic wave impedance inversion device is provided, comprising:

[0045] Determine the unit for well seismic calibration and determine the target layer section sample;

[0046] An acquisition unit, used to acquire a density curve and a velocity curve of the logging data, and to extract training sample labels from the density curve and the velocity curve of the logging data;

[0047] The first construction unit is used to obtain seismic data of a layer segment with the same depth as the target layer segment from the seismic data, and perform spectrum analysis on the obtained seismic data, and use the seismic data after the spectrum analysis and the density data to construct a training data set for a density interpretation model, and use the seismic data after the spectrum analysis and the velocity data to construct a training data set for a velocity interpretation model;

[0048] The second construction unit is used to construct a velocity interpretation model and a density interpretation model based on the training data set, respectively, and map the seismic data to velocity and density according to the velocity interpretation model and the density interpretation model; wherein the velocity interpretation model and the density interpretation model are both constructed using gated recurrent GRU neural units, and the number of GRU layers is 3; the dimensions of the seismic data input to the velocity interpretation model and the density interpretation model are (B, 50, m), and the dimensions of the output velocity and density are both (B, 1, m), and B represents the number of samples of the model during each training;

[0049] A first training unit is used to train a density explanation model using the training data in the training data set of the density explanation model as a final density explanation model P1 after training is completed, and to train a speed explanation model using the training data in the training data set of the speed explanation model as a final speed explanation model S1 after training is completed;

[0050] The third construction unit is used to migrate the model parameters of the model P1 and the model S1 respectively, so as to construct a wave impedance inversion model of multi-task learning, wherein the wave impedance inversion model is a parallel connection of the density interpretation model and the velocity interpretation model;

[0051] The second training unit is used to conduct comprehensive training on the wave impedance inversion model and determine its loss function;

[0052] The inversion processing unit is used to use the wave impedance inversion model to complete the inversion processing of seismic data on density, velocity and wave impedance respectively.

[0053] In one embodiment, the device further comprises:

[0054] The calculation unit is used to obtain velocity data and density data according to the density curve and velocity curve of the logging data, and calculate the product of velocity and density as wave impedance.

[0055] In one possible implementation manner, the acquisition unit is further used for:

[0056] Use the well coordinates to obtain the seismic trace data in the unit area grid centered at the corresponding position as the model input data, and use the following spectrum analysis method to obtain the frequency domain characteristics of the seismic data;

[0057]

[0058] in:

[0059] is a one-dimensional seismic trace signal, is the spectrum of the one-dimensional seismic signal in the frequency domain, is a plural unit, is the fundamental frequency of the one-dimensional seismic trace signal;

[0060] According to the spectrum characteristics, select the first main frequency and use the following bandpass filter , get the seismic waveform corresponding to the first main frequency:

[0061]

[0062] in, is the first main frequency, is the low-pass frequency value centered on the first main frequency, It is the high-pass frequency value centered on the first main frequency.

[0063] In one embodiment, the second construction unit is further used for:

[0064] Determine the number of target layer sample points as m, extract density and velocity training sample labels from the well logging data, the label dimension is (1, m), and obtain p×p grid seismic data of the same depth layer as the target layer from the seismic data, the dimension of which is (p, p, m);

[0065] The dimension of wave impedance is (1, m), and the seismic waveform corresponding to the first main frequency of seismic data is obtained by frequency filtering, and its dimension is (p, p, m);

[0066] The seismic data and the seismic waveform corresponding to the main frequency f are merged, the seismic data at different grid positions are merged, and the merged data is used as the data of the input model, and its dimension is (50, m).

[0067] The operation process of each GRU layer in the three-layer GRU of the speed interpretation module and the density interpretation module includes: controlling how the state of the previous moment affects the current state; controlling how the current input affects the state of the previous moment; calculating the candidate hidden state according to the current input and the reset state of the previous moment; and calculating the final hidden state at the current moment according to the influence of the state of the previous moment on the current state.

[0068] In one embodiment, the loss function of the wave impedance inversion model includes three parts: density loss , speed loss and wave impedance loss , calculated as follows:

[0069]

[0070]

[0071]

[0072]

[0073] in:

[0074] is the inversion loss of the density module;

[0075] is the inversion loss of the velocity module;

[0076] is the inversion loss of the wave impedance module;

[0077] N : The number of training samples;

[0078] : represents the true density value;

[0079] : represents the density value inverted by the density interpretation module;

[0080] : Indicates the true speed value;

[0081] : represents the velocity value inverted by the velocity interpretation module;

[0082] : represents the real wave impedance value;

[0083] : Represents the inversion value of the seismic wave impedance inversion model.

[0084] The technical solution provided by the embodiments of the present application may have the following beneficial effects:

[0085] This application uses well seismic calibration to determine the seismic data of the target layer sample, uses gated recurrent GRU neural units to build velocity and density interpretation models, and constructs a wave impedance inversion model. Therefore, the wave impedance results obtained based on the relevant model training simulation are more accurate, and the geophysical prior knowledge is effectively integrated into the inversion model, which improves the accuracy of seismic wave impedance inversion, is more robust, and improves the vertical resolution.

[0086] It should be understood that the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the present application. BRIEF DESCRIPTION OF THE DRAWINGS

[0087] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the present application and, together with the description, serve to explain the principles of the present application.

[0088] Figure 1 A schematic diagram of a flow chart of a multi-task intelligent seismic wave impedance inversion method according to an embodiment of the present application;

[0089] Figure 2 A schematic diagram of seismic one-dimensional waveform data and wave impedance data shown in an embodiment of the present application;

[0090] Figure 3 A schematic diagram of a seismic wave impedance inversion model framework shown in an embodiment of the present application;

[0091] Figure 4 A schematic diagram of the composition structure of a multi-task intelligent seismic wave impedance inversion device shown in an embodiment of the present application;

[0092] Figure 5 It is a schematic structural diagram of an electronic device according to an embodiment of the present application. DETAILED DESCRIPTION

[0093] Exemplary embodiments will be described in detail herein, examples of which are shown in the accompanying drawings. When the following description refers to the drawings, the same numbers in different drawings represent the same or similar elements unless otherwise indicated. The implementations described in the following exemplary embodiments do not represent all implementations consistent with the present application. Instead, they are merely examples of devices and methods consistent with some aspects of the present application as detailed in the appended claims.

[0094] Figure 1 FIG. 1 is a flow chart of a multi-task intelligent seismic wave impedance inversion method according to an embodiment of the present application. Figure 1 As shown, the multi-task intelligent seismic wave impedance inversion method of the embodiment of the present application includes the following steps:

[0095] Step 101, perform well seismic calibration and determine target layer samples.

[0096] Seismic reservoir prediction is becoming more and more important in the exploration and development of oil and gas reservoirs. Well seismic calibration, as a key technology for structural interpretation of seismic data and seismic reservoir prediction, is a bridge between geology, logging, and seismic. It associates seismic data with logging data in the depth domain, and combines logging data with seismic data with lateral distribution advantage characteristics. The quality of well seismic calibration directly affects the effectiveness and accuracy of structural interpretation of seismic data and seismic reservoir prediction. The embodiment of the present application requires accurate calibration of relevant parameters of well seismic, such as seismic data and logging data, and at the same time determines the sample of the target layer segment to be used for seismic reservoir prediction.

[0097] Step 102: Obtain a density curve and a velocity curve of the logging data, and extract training sample labels from the density curve and the velocity curve of the logging data.

[0098] In the embodiment of the present application, training sample labels can be intercepted from the density curve and velocity curve of the logging data, and seismic data of the same depth segment as the target layer segment can be obtained from the seismic data in a 5×5 grid, and spectral analysis can be performed. Seismic data and density data are used to establish a density interpretation model training data set, and seismic data and velocity data are used to establish a velocity interpretation model training data set. The size of the grid can be set as needed, for example, it can also be a 10×10 grid, a 15×15 grid, or a 20×20 grid.

[0099] Step 103, obtaining seismic data of the same depth segment as the target segment from the seismic data, and performing spectral analysis on the obtained seismic data, using the seismic data after spectral analysis and density data to construct a training data set for a density interpretation model, and using the seismic data after spectral analysis and velocity data to construct a training data set for a velocity interpretation model.

[0100] In the embodiment of the present application, after the well seismic calibration, the number of sample points of the target layer is determined to be m, and the density and velocity training sample labels are intercepted from the well logging data, and their dimensions are (1, m). The seismic data of the 5×5 grid of the same depth segment as the target layer is obtained from the seismic data, and its dimensions are (5, 5, m). The wave impedance data is calculated by multiplying the density and velocity, and its dimensions are (1, m). The seismic waveform corresponding to the first main frequency of the seismic data is obtained by frequency filtering, and its dimensions are (5, 5, m).

[0101] Among them, the wave impedance ;

[0102] is the wave impedance, is the density, It's speed.

[0103] Using the well coordinates, the seismic trace data in a 5×5 grid centered at the corresponding position is obtained as the model input data, and the frequency domain characteristics of the seismic data are obtained using the spectrum analysis method (such as formula 1);

[0104]

[0105] in:

[0106] : is a one-dimensional seismic trace signal.

[0107] : is the frequency spectrum of the one-dimensional seismic signal in the frequency domain.

[0108] : plural unit.

[0109] : The fundamental frequency of one-dimensional seismic trace signal.

[0110] Finally, according to the spectrum characteristics, the first main frequency is selected, and then a bandpass filter is used to obtain the seismic waveform corresponding to the first main frequency. , as follows:

[0111]

[0112] in:

[0113] is the first main frequency, is the low-pass frequency value centered on the first main frequency, It is the high-pass frequency value centered on the first main frequency.

[0114] Finally, the seismic data and the seismic waveform corresponding to the main frequency f are merged, and the seismic data at different grid locations are merged to form the input model data, i.e., the training data set, whose dimension is (50, m).

[0115] Step 104: construct a velocity interpretation model and a density interpretation model based on the training data set, and map the seismic data into velocity and density according to the velocity interpretation model and the density interpretation model.

[0116] In the embodiment of the present application, the velocity interpretation model and the density interpretation model are both constructed using gated recurrent GRU neural units, and the number of GRU layers is 3; the dimension of the seismic data input to the velocity interpretation model and the density interpretation model is (B, 50, m), and the dimension of the output velocity and density is (B, 1, m), where B represents the number of samples of the model to be trained each time.

[0117] Among them, the operation process of each layer of GRU in the three layers of GRU of the speed interpretation module and the density interpretation module includes: controlling how the state of the previous moment affects the current state; controlling how the current input affects the state of the previous moment; calculating the candidate hidden state according to the current input and the reset state of the previous moment; and calculating the final hidden state at the current moment according to the influence of the state of the previous moment on the current state.

[0118] Step 105, using the training data in the training data set of the density explanation model to train the density explanation model as the final density explanation model P1 after the training is completed, and using the training data in the training data set of the speed explanation model to train the speed explanation model as the final speed explanation model S1 after the training is completed.

[0119] Step 106 , the model parameters of the model P1 and the model S1 are migrated respectively, so as to construct a wave impedance inversion model of multi-task learning.

[0120] In the embodiment of the present application, the wave impedance inversion model is a parallel combination of the density interpretation model and the velocity interpretation model. Specifically, the loss function of the wave impedance inversion model includes three parts: density loss , speed loss and wave impedance loss , calculated as follows:

[0121]

[0122]

[0123]

[0124]

[0125] in:

[0126] is the inversion loss of the density module;

[0127] is the inversion loss of the velocity module;

[0128] is the inversion loss of the wave impedance module;

[0129] N : The number of training samples;

[0130] : represents the true density value;

[0131] : represents the density value inverted by the density interpretation module;

[0132] : Indicates the true speed value;

[0133] : represents the velocity value inverted by the velocity interpretation module;

[0134] : represents the real wave impedance value;

[0135] : Represents the inversion value of the seismic wave impedance inversion model.

[0136] Step 107, comprehensively train the wave impedance inversion model and determine its loss function; use the wave impedance inversion model to complete the inversion processing of density, velocity and wave impedance of seismic data respectively.

[0137] The last layer of the wave impedance inversion model is the vector product operator. The wave impedance is directly calculated using the output of the velocity interpretation module and the output of the density interpretation module. The input of the wave impedance inversion model is seismic data, whose dimension is (B, 50, m). The output of the wave impedance inversion model contains three parts, namely velocity, density and wave impedance, whose dimension is (B, 1, m).

[0138] The essence of the technical solution of the embodiment of the present application is further illustrated by specific examples below.

[0139] Through the above-mentioned technical means, the technical solution of the embodiment of the present application realizes high-precision seismic wave impedance inversion.

[0140] The essence of the technical solution of the embodiment of the present application is further illustrated by specific examples below.

[0141] The Lvrong (LU) interval of the M oilfield was used as the research target. Figure 2 This is a schematic diagram of seismic one-dimensional waveform data and wave impedance data shown in an embodiment of the present application, such as Figure 2 As shown, the data required in the implementation process include 1D seismic data, 1D density data in well logging data, 1D velocity data and 1D wave impedance data.

[0142] The specific implementation of the technical solution of the embodiment of the present application is as follows:

[0143] Step (1), after well seismic calibration, select the target layer LU sample thickness of 40 meters, the number of sample points in the depth direction of 80, and extract the training sample labels from the density curve and velocity curve of the logging data. Obtain the seismic data of the same depth segment from the seismic data in a 5×5 grid and perform spectrum analysis to determine that the main seismic frequency f of the LU layer segment is 45HZ. Use seismic data and density data to establish a density interpretation model training data set, use seismic data and velocity data to establish a velocity interpretation model training data set, and use seismic data and wave impedance data to establish a wave impedance inversion data set. The input data dimension of the wave impedance inversion model is (32, 50, 80), and the output dimension is (32, 1, 80).

[0144] Step (2), construct velocity interpretation model and density interpretation model to complete the nonlinear mapping from seismic data to velocity and density. Both velocity interpretation model and density interpretation model are constructed using GRU neural units, and the number of GRU neural units is 3 layers. The last layer of both velocity interpretation model and density interpretation model uses linear rectification function (Relu) activation function. The input dimensions of velocity interpretation model and density interpretation model are both (32, 50, 80) seismic data. The dimensions of output velocity and density are (32, 1, 80).

[0145] Step (3): First, use the seismic data and density data to train the density interpretation model, and save the final model P1 after training. Then use the seismic data and velocity data to train the velocity interpretation model, and save the final model S1 after training. The hyperparameters used when training the velocity model and the density model independently are: the learning rate is set to 0.001, the epoch (the process in which the entire data set is completely traversed and learned by the model once) is set to 1000, the optimization function is set to the Adam (nonlinear activation function used in neural networks) optimization function, the batch (a group of samples input into the network at the same time when training the neural network) is set to 32, and the feature dimension channel of the model is 50 for the first layer, 128 for the second layer, and 256 for the third layer.

[0146] Step (4), migrate the model parameters of velocity model P1 and density model S1, and construct a multi-task learning wave impedance inversion model. The wave impedance inversion model is a parallel combination of the density interpretation model and the velocity interpretation model. The last layer of the wave impedance inversion model is a vector product operator. Use the output of the velocity interpretation module and the output of the density interpretation module to directly calculate the wave impedance.

[0147] Wave impedance

[0148] is the wave impedance, is the density, It's speed.

[0149] like Figure 3 As shown, the input of the wave impedance inversion model of the embodiment of the present application is seismic data, whose dimension is (32, 50, 80), and the output of the wave impedance inversion model includes three parts, namely velocity, density and wave impedance, whose dimension is (32, 1, 80). During comprehensive training, the loss function includes three parts, velocity explanation loss, density explanation loss, and wave impedance explanation loss. The hyperparameters used in the comprehensive training of the wave impedance inversion model are learning rate set to 0.00001, epoch set to 5000, optimization function set to Adam optimization function, and batch set to 32.

[0150] After completing the training of the seismic inversion model, the model can be directly used in the non-well seismic data of the bottom section to be processed. The wave impedance inversion of all seismic traces is completed cyclically, and finally the wave impedance data corresponding to the seismic traces are combined according to the position of the seismic traces to form a wave impedance body.

[0151] Figure 4 This is a schematic diagram of the composition structure of a multi-task intelligent seismic wave impedance inversion device according to an embodiment of the present application. Figure 4As shown, the multi-task intelligent seismic wave impedance inversion device of the embodiment of the present disclosure includes:

[0152] A first determination unit 40 is used to perform well seismic calibration and determine target layer section samples;

[0153] An acquisition unit 41 is used to acquire a density curve and a velocity curve of the well logging data, and to extract training sample labels from the density curve and the velocity curve of the well logging data;

[0154] A first construction unit 42 is used to obtain seismic data of a layer segment with the same depth as the target layer segment from the seismic data, and perform spectrum analysis on the obtained seismic data, and use the seismic data after the spectrum analysis and the density data to construct a training data set for a density interpretation model, and use the seismic data after the spectrum analysis and the velocity data to construct a training data set for a velocity interpretation model;

[0155] The second construction unit 43 is used to construct a velocity interpretation model and a density interpretation model based on the training data set, respectively, and map the seismic data to velocity and density according to the velocity interpretation model and the density interpretation model respectively; wherein the velocity interpretation model and the density interpretation model are both constructed by using gated recurrent GRU neural units, and the number of GRU layers is 3; the dimensions of the seismic data input to the velocity interpretation model and the density interpretation model are (B, 50, m), and the dimensions of the output velocity and density are both (B, 1, m), and B represents the number of samples of the model during each training;

[0156] A first training unit 44 is used to train a density explanation model using the training data in the training data set of the density explanation model as a final density explanation model P1 after training, and to train a speed explanation model using the training data in the training data set of the speed explanation model as a final speed explanation model S1 after training;

[0157] The third construction unit 45 is used to migrate the model parameters of the model P1 and the model S1 respectively, so as to construct a wave impedance inversion model of multi-task learning, wherein the wave impedance inversion model is a parallel connection of the density interpretation model and the velocity interpretation model;

[0158] The second training unit 46 is used to perform comprehensive training on the wave impedance inversion model and determine its loss function;

[0159] The inversion processing unit 47 is used to use the wave impedance inversion model to complete the inversion processing of the seismic data on density, velocity and wave impedance respectively.

[0160] exist Figure 4 Based on the multi-task intelligent seismic wave impedance inversion device shown, the multi-task intelligent seismic wave impedance inversion device of the embodiment of the present application further includes:

[0161] Computational Unit ( Figure 4 ), which is used to obtain velocity data and density data according to the density curve and velocity curve of the logging data, respectively, and calculate the product of velocity and density as the wave impedance.

[0162] In one possible implementation, the acquisition unit 41 is further configured to:

[0163] Use the well coordinates to obtain the seismic trace data in the unit area grid centered at the corresponding position as the model input data, and use the following spectrum analysis method to obtain the frequency domain characteristics of the seismic data;

[0164]

[0165] in:

[0166] is a one-dimensional seismic trace signal, is the spectrum of the one-dimensional seismic signal in the frequency domain, is a plural unit, is the fundamental frequency of the one-dimensional seismic trace signal;

[0167] According to the spectrum characteristics, select the first main frequency and use the following bandpass filter , get the seismic waveform corresponding to the first main frequency:

[0168]

[0169] in, is the first main frequency, is the low-pass frequency value centered on the first main frequency, It is the high-pass frequency value centered on the first main frequency.

[0170] In one embodiment, the second constructing unit 43 is further used for:

[0171] Determine the number of target layer sample points as m, extract density and velocity training sample labels from the well logging data, the label dimension is (1, m), and obtain p×p grid seismic data of the same depth layer as the target layer from the seismic data, the dimension of which is (p, p, m);

[0172] The dimension of wave impedance is (1, m), and the seismic waveform corresponding to the first main frequency of seismic data is obtained by frequency filtering, and its dimension is (p, p, m);

[0173] The seismic data and the seismic waveform corresponding to the main frequency f are merged, the seismic data at different grid positions are merged, and the merged data is used as the data of the input model, and its dimension is (50, m).

[0174] The operation process of each GRU layer in the three-layer GRU of the speed interpretation module and the density interpretation module includes: controlling how the state of the previous moment affects the current state; controlling how the current input affects the state of the previous moment; calculating the candidate hidden state according to the current input and the reset state of the previous moment; and calculating the final hidden state at the current moment according to the influence of the state of the previous moment on the current state.

[0175] In one embodiment, the loss function of the wave impedance inversion model includes three parts: density loss , speed loss and wave impedance loss , calculated as follows:

[0176]

[0177]

[0178]

[0179]

[0180] in:

[0181] is the inversion loss of the density module;

[0182] is the inversion loss of the velocity module;

[0183] is the inversion loss of the wave impedance module;

[0184] N : The number of training samples;

[0185] : represents the true density value;

[0186] : represents the density value inverted by the density interpretation module;

[0187] : Indicates the true speed value;

[0188] : represents the velocity value inverted by the velocity interpretation module;

[0189] : represents the real wave impedance value;

[0190] : Represents the inversion value of the seismic wave impedance inversion model.

[0191] In an exemplary embodiment, the aforementioned units and the like may be implemented by one or more central processing units (CPU), graphics processing units (GPU), application specific integrated circuits (ASIC), DSPs, programmable logic devices (PLD), complex programmable logic devices (CPLD), field programmable gate arrays (FPGA), general-purpose processors, controllers, microcontrollers (MCU), microprocessors, or other electronic components.

[0192] Regarding the device in the above embodiment, the specific manner in which each module and unit performs operations has been described in detail in the embodiment of the method, and will not be elaborated here.

[0193] Figure 5 8 is a schematic block diagram of an example network element 800 that can be used to implement an embodiment of the present disclosure. Figure 5 As shown, the network element 800 includes a computing unit 801, which can perform various appropriate actions and processes according to a computer program stored in a read-only memory (ROM) 802 or a computer program loaded from a storage unit 808 to a random access memory (RAM) 803. In the RAM 803, various programs and data required for the operation of the network element 800 can also be stored. The computing unit 801, the ROM 802, and the RAM 803 are connected to each other via a bus 804. An input / output (I / O) interface 805 is also connected to the bus 804.

[0194] Multiple components in the network element 800 are connected to the I / O interface 805, including: an input unit 806, such as a keyboard, a mouse, etc.; an output unit 807, such as various types of displays, speakers, etc.; a storage unit 808, such as a disk, an optical disk, etc.; and a communication unit 809, such as a network card, a modem, a data processing transceiver, etc. The communication unit 809 allows the network element 800 to exchange information / data with other devices through a computer network such as the Internet and / or various telecommunication networks.

[0195] The computing unit 801 may be a variety of general and / or special processing components with processing and computing capabilities. Some examples of the computing unit 801 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various dedicated artificial intelligence (AI) computing chips, various computing units running machine learning model algorithms, digital signal processors (DSPs), and any appropriate processors, controllers, microcontrollers, etc. The computing unit 801 performs the various methods and processes described above, such as the multi-task intelligent seismic wave impedance inversion method. For example, in some embodiments, the multi-task intelligent seismic wave impedance inversion method may be implemented as a computer software program, which is tangibly included in a machine-readable medium, such as a storage unit 808. In some embodiments, part or all of the computer program may be loaded and / or installed on the network element 800 via the ROM 802 and / or the communication unit 809. When the computer program is loaded into the RAM 803 and executed by the computing unit 801, one or more steps of the multi-task intelligent seismic wave impedance inversion method described above may be performed. Alternatively, in other embodiments, the computing unit 801 may be configured to execute the steps of the multi-task intelligent seismic wave impedance inversion method in any other appropriate manner (eg, by means of firmware).

[0196] Various implementations of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard products (ASSPs), systems on chips (SOCs), complex programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various implementations can include: being implemented in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which can be a special purpose or general purpose programmable processor that can receive data and instructions from a storage system, at least one input device, and at least one output device, and transmit data and instructions to the storage system, the at least one input device, and the at least one output device.

[0197] The program code for implementing the method of the present disclosure may be written in any combination of one or more programming languages. These program codes may be provided to a processor or controller of a general-purpose computer, a special-purpose computer, or other programmable data processing device, so that the program code, when executed by the processor or controller, enables the functions / operations specified in the flow chart and / or block diagram to be implemented. The program code may be executed entirely on the machine, partially on the machine, partially on the machine and partially on a remote machine as a stand-alone software package, or entirely on a remote machine or server.

[0198] In the context of the present disclosure, a machine-readable medium may be a tangible medium that may contain or store a program for use by or in conjunction with an instruction execution system, device, or equipment. A machine-readable medium may be a machine-readable signal medium or a machine-readable storage medium. A machine-readable medium may include, but is not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, device, or device, or any suitable combination of the foregoing. A more specific example of a machine-readable storage medium may include an electrical connection based on one or more lines, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

[0199] To provide interaction with a user, the systems and techniques described herein can be implemented on a computer having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and pointing device (e.g., a mouse or trackball) through which the user can provide input to the computer. Other types of devices can also be used to provide interaction with the user; for example, the feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including acoustic input, voice input, or tactile input).

[0200] The systems and techniques described herein may be implemented in a computing system that includes back-end components (e.g., as a data server), or a computing system that includes middleware components (e.g., an application server), or a computing system that includes front-end components (e.g., a user computer with a graphical user interface or a web browser through which a user can interact with implementations of the systems and techniques described herein), or a computing system that includes any combination of such back-end components, middleware components, or front-end components. The components of the system may be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include: a local area network (LAN), a wide area network (WAN), and the Internet.

[0201] A computer system may include a client and a server. The client and the server are generally remote from each other and usually interact through a communication network. The relationship of client and server is generated by computer programs running on respective computers and having a client-server relationship with each other. The server may be a cloud server, a server of a distributed system, or a server combined with a blockchain.

[0202] It should be understood that the various forms of processes shown above can be used to reorder, add or delete steps. For example, the steps recorded in this disclosure can be executed in parallel, sequentially or in different orders, as long as the desired results of the technical solutions disclosed in this disclosure can be achieved, and this document does not limit this.

[0203] In addition, the terms "first" and "second" are used for descriptive purposes only and should not be understood as indicating or implying relative importance or implicitly indicating the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include at least one of the features. In the description of the present disclosure, the meaning of "plurality" is two or more, unless otherwise clearly and specifically defined.

[0204] The above is only a specific embodiment of the present disclosure, but the protection scope of the present disclosure is not limited thereto. Any person skilled in the art who is familiar with the technical field can easily think of changes or substitutions within the technical scope disclosed in the present disclosure, which should be included in the protection scope of the present disclosure. Therefore, the protection scope of the present disclosure should be based on the protection scope of the claims.

Claims

1. A multi-task intelligent seismic wave impedance inversion method, characterized in that: The method comprises: Conduct well seismic calibration and determine target layer samples; Obtaining a density curve and a velocity curve of the well logging data, and extracting training sample labels from the density curve and the velocity curve of the well logging data; Acquire seismic data of a layer segment at the same depth as the target layer segment from the seismic data, perform spectrum analysis on the acquired seismic data, use the seismic data after spectrum analysis and density data to construct a training data set for a density interpretation model, and use the seismic data after spectrum analysis and velocity data to construct a training data set for a velocity interpretation model; A velocity interpretation model and a density interpretation model are respectively constructed based on the training data set, and the seismic data are respectively mapped to velocity and density according to the velocity interpretation model and the density interpretation model; wherein the velocity interpretation model and the density interpretation model are both constructed using gated recurrent GRU neural units, and the number of GRU layers is 3; the dimension of the seismic data input to the velocity interpretation model and the density interpretation model is (B, 50, m), and the dimension of the output velocity and density is (B, 1, m), and B represents the number of samples of the model at each training; The density explanation model is trained using the training data in the training data set of the density explanation model, as the final density explanation model P1 after the training is completed; the speed explanation model is trained using the training data in the training data set of the speed explanation model, as the final speed explanation model S1 after the training is completed; The model parameters of model P1 and model S1 are migrated respectively, so as to construct a wave impedance inversion model of multi-task learning, wherein the wave impedance inversion model is a parallel combination of the density interpretation model and the velocity interpretation model; The wave impedance inversion model is comprehensively trained and its loss function is determined; the wave impedance inversion model is used to complete the inversion processing of seismic data on density, velocity and wave impedance respectively.

2. The method according to claim 1, characterized in that The method further comprises: The velocity data and density data are obtained respectively according to the density curve and velocity curve of the logging data, and the product of velocity and density is calculated as the wave impedance.

3. The method according to claim 1 or 2, characterized in that: The step of acquiring seismic data of a layer segment with the same depth as the target layer segment from the seismic data and performing spectrum analysis on the acquired seismic data comprises: Use the well coordinates to obtain the seismic trace data in the unit area grid centered at the corresponding position as the model input data, and use the following spectrum analysis method to obtain the frequency domain characteristics of the seismic data; in: is a one-dimensional seismic trace signal, is the spectrum of the one-dimensional seismic signal in the frequency domain, is a plural unit, is the fundamental frequency of the one-dimensional seismic trace signal; According to the spectrum characteristics, select the first main frequency and use the following bandpass filter , get the seismic waveform corresponding to the first main frequency: in, is the first main frequency, is the low-pass frequency value centered on the first main frequency, It is the high-pass frequency value centered on the first main frequency.

4. The method according to claim 3, characterized in that The constructing of a velocity explanation model and a density explanation model based on the training data set respectively includes: Determine the number of target layer sample points as m, extract density and velocity training sample labels from the well logging data, the label dimension is (1, m), and obtain p×p grid seismic data of the same depth layer as the target layer from the seismic data, the dimension of which is (p, p, m); The dimension of wave impedance is (1, m), and the seismic waveform corresponding to the first main frequency of seismic data is obtained by frequency filtering, and its dimension is (p, p, m); The seismic data and the seismic waveform corresponding to the main frequency f are merged, the seismic data at different grid positions are merged, and the merged data is used as the data of the input model, and its dimension is (50, m).

5. The method according to claim 1, characterized in that The operation process of each layer of GRU in the three layers of GRU of the speed explanation model and the density explanation model includes: controlling how the state of the previous moment affects the current state; controlling how the current input affects the state of the previous moment; calculating the candidate hidden state according to the current input and the reset state of the previous moment; and calculating the final hidden state at the current moment according to the influence of the state of the previous moment on the current state.

6. The method according to claim 1, characterized in that The loss function of the wave impedance inversion model consists of three parts: density loss , speed loss and wave impedance loss , calculated as follows: in: is the inversion loss of the density module; is the inversion loss of the velocity module; is the inversion loss of the wave impedance module; N : The number of training samples; : represents the true density value; : represents the density value inverted by the density interpretation module; : Indicates the true speed value; : represents the velocity value inverted by the velocity interpretation module; : represents the real wave impedance value; : Represents the inversion value of the seismic wave impedance inversion model.

7. A multi-task intelligent seismic wave impedance inversion device, characterized in that: The device comprises: The first determination unit is used to perform well seismic calibration and determine target layer section samples; An acquisition unit, used to acquire a density curve and a velocity curve of the logging data, and to extract training sample labels from the density curve and the velocity curve of the logging data; The first construction unit is used to obtain seismic data of a layer segment with the same depth as the target layer segment from the seismic data, and perform spectrum analysis on the obtained seismic data, and use the seismic data after the spectrum analysis and the density data to construct a training data set for a density interpretation model, and use the seismic data after the spectrum analysis and the velocity data to construct a training data set for a velocity interpretation model; The second construction unit is used to construct a velocity interpretation model and a density interpretation model based on the training data set, respectively, and map the seismic data to velocity and density according to the velocity interpretation model and the density interpretation model; wherein the velocity interpretation model and the density interpretation model are both constructed using gated recurrent GRU neural units, and the number of GRU layers is 3; the dimensions of the seismic data input to the velocity interpretation model and the density interpretation model are (B, 50, m), and the dimensions of the output velocity and density are both (B, 1, m), and B represents the number of samples of the model during each training; A first training unit is used to train a density explanation model using the training data in the training data set of the density explanation model as a final density explanation model P1 after training is completed, and to train a speed explanation model using the training data in the training data set of the speed explanation model as a final speed explanation model S1 after training is completed; The third construction unit is used to migrate the model parameters of the model P1 and the model S1 respectively, so as to construct a wave impedance inversion model of multi-task learning, wherein the wave impedance inversion model is a parallel connection of the density interpretation model and the velocity interpretation model; The second training unit is used to conduct comprehensive training on the wave impedance inversion model and determine its loss function; The inversion processing unit is used to use the wave impedance inversion model to complete the inversion processing of seismic data on density, velocity and wave impedance respectively.

8. The device according to claim 7, characterized in that The device also includes: The calculation unit is used to obtain velocity data and density data according to the density curve and velocity curve of the logging data, and calculate the product of velocity and density as wave impedance.

9. The device according to claim 7 or 8, characterized in that The acquisition unit is further used for: Use the well coordinates to obtain the seismic trace data in the unit area grid centered at the corresponding position as the model input data, and use the following spectrum analysis method to obtain the frequency domain characteristics of the seismic data; in: is a one-dimensional seismic trace signal, is the spectrum of the one-dimensional seismic signal in the frequency domain, is a plural unit, is the fundamental frequency of the one-dimensional seismic trace signal; According to the spectrum characteristics, select the first main frequency and use the following bandpass filter , get the seismic waveform corresponding to the first main frequency: in, is the first main frequency, is the low-pass frequency value centered on the first main frequency, It is the high-pass frequency value centered on the first main frequency.

10. The device according to claim 9, characterized in that The second building block is further used for: Determine the number of target layer sample points as m, extract density and velocity training sample labels from the well logging data, the label dimension is (1, m), and obtain p×p grid seismic data of the same depth layer as the target layer from the seismic data, the dimension of which is (p, p, m); The dimension of wave impedance is (1, m), and the seismic waveform corresponding to the first main frequency of seismic data is obtained by frequency filtering, and its dimension is (p, p, m); The seismic data and the seismic waveform corresponding to the main frequency f are merged, the seismic data at different grid positions are merged, and the merged data is used as the data of the input model, and its dimension is (50, m).

Images